CN101141230A

CN101141230A - Method and communications device for adapting the data transfer rate in a communications device

Info

Publication number: CN101141230A
Application number: CNA2007101496150A
Authority: CN
Inventors: M·德特林; B·拉夫
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 2002-04-08
Filing date: 2003-04-01
Publication date: 2008-03-12
Anticipated expiration: 2023-04-01
Also published as: KR20100072342A; EP1497943B1; AU2003232589A1; CN1647435A; CN100347981C; EP1497943A2; KR20040095366A; BRPI0318842B1; CN101141230B; BR0309059A; JP4847679B2; US8009550B2; US20050141549A1; KR101045449B1; JP2005529509A; BRPI0309059B1; KR101120368B1

Abstract

A method is provided for adapting the data transfer rate of a data flow in a communication device according to which: the data flow can be subdivided into at least one data block containing transmission bits to be transmitted; the transmission bits are formed by a coding process from information-carrying input bits; transmission bits determined from a data block of the data flow are removed (punctured) in order to adapt the data transfer rate; a puncturing pattern stipulates which transmission bits are to be removed, and; the puncturing pattern is constructed in such a manner that transmission bits are preferably removed that, during the coding process, depend on few input bits. The present invention also relates to a corresponding communication device.

Description

Method for matching data rates in a communication device and communication device

This application is a divisional application of the following invention patent international applications, international application numbers, which have been filed on 1/4/2003: PCT/DE03/01061, national application number: 03807836.8, invention name: "method of matching data rates in communication device and communication device".

Technical Field

The present invention relates to a method for adapting a data rate in a communication device, and to a corresponding communication device.

Background

Different applications in a communication system operate with mostly different data rates. However, most systems provide only a fixed or raw (Roh) data rate, or only a discrete set of such data rates, based on the transmission channel, for example, by being incorporated into certain transmission formats. It is generally necessary to match the data rates to each other on the respective interfaces. This is illustrated below with an example from UMTS standardization:

currently, work is being carried out on the standardization of the so-called UMTS mobile radio standard ('universal mobile telecommunications system') for mobile radio communication devices of the third mobile radio generation. The current state of the UMTS standardization specifies that data to be transmitted via high-frequency channels should be subjected to channel coding, a convolutional code being used for this purpose in particular. The data to be transmitted are redundantly coded by channel coding, so that the transmitted data can be recovered more reliably at the receiver. The codes respectively used in channel coding are characterized by their code rate r = k/n, where k denotes the number of data bits or message bits to be transmitted and n denotes the number of bits present after coding. The smaller the code rate, the more efficient the code is generally. The problem connected with coding is however that the data rate is reduced by a factor r.

In order to match the data rate of the encoded data stream to the respective possible transmission rate, rate matching is implemented in the transmitter, wherein either bits are removed from the data stream or bits are doubled in the data stream in a certain pattern (Muster). The removed bits are called 'dotting' ('Punktieren') and doubling is called 'repetition'.

According to the current state of UMTS standardization, an algorithm is proposed for rate matching which uses an approximately regular dotting pattern to implement dotting, i.e. the bits to be dotted are distributed equidistantly over the respective data word to be dotted.

In addition, it is disclosed that the Bit Error Rate (BER) decreases at the edges of the respectively coded data word during the convolutional coding. It is also disclosed that the bit error rate within a data word can be locally changed by unevenly distributed dotting. It is also known from WO 01/26273A1 and WO 01/39421A1 that it is advantageous to dot the individual data words of the data stream according to a dotting pattern for matching the data rate, wherein the dotting pattern is formed such that it has a continuously increasing dotting rate (Punktierungsrate) from the center of the individual data words in the direction of at least one end of the individual data words.

Disclosure of Invention

It is therefore an object of the present invention to provide a method for adapting the data rate of a data stream in a communication device, and a corresponding communication device, which ensure a satisfactory bit error rate, in particular for use in mobile radio systems with convolutional coding.

This object is achieved according to the invention by a method for adapting the data rate of a data stream in a communication device or by a communication device. Preferred and advantageous embodiments of the invention are also specified.

In the method: said data stream can be divided into at least one data word group containing transmission bits to be transmitted, said transmission bits being formed from the information-carrying input bits by means of an encoding process, some transmission bits being removed from the data word group of said data stream in order to match said data rate, being predefined by means of a dotting pattern: which transmission bits should be removed and the dotting pattern is designed such that: preferably, the transmitted bits associated with a small number of input bits by said encoding process are removed, and said dotting pattern is formed by: the cumulative dotting intensity is found, which accounts for: by removing transmission bits from said data word, which component of the information bits has been removed, a decision function is formed depending on said accumulated dotting strength such that said decision function for finding said dotting pattern is minimized.

In the method: in the method, in order to match the data rate, certain transmission bits are removed from the data word set of the data stream, which transmission bits are to be removed are predetermined by a dotting pattern, and the dotting pattern is designed in such a way that: preferably, the transmission bits associated with a small number of input bits by said encoding process are removed, and said dotting pattern is constructed such that: viewed from the end of the front of the data word to be dotted, a segment from the following sequence is contained by the dotted line: 1,4,2,3,8,7,5,6, 15, 12, 14, 11, 10,9, wherein "1" corresponds to said first bit position.

In the method: in the method, in order to match the data rate, some transmission bits are removed from the data word groups of the data stream, which are predefined by a dotting pattern: which transmission bits should be removed and the dotting pattern is designed such that: the transmission bits associated with a small number of input bits by the coding process are preferably removed, and the dotting pattern is constructed such that: viewed from the rear end of the data word to be doted, a segment from the following sequence is contained by the dotting: 0,4,6,1,2, 15, 12, 10,9,7,4,5, 18, 13,8, wherein "0" corresponds to said last bit position.

In the method: said data stream can be divided into at least one data word group containing transmission bits to be transmitted, said transmission bits being formed from information-carrying input bits by means of an encoding process,

in the method, in order to match the data rate, certain transmission bits are repeated from groups of data words of the data stream, a repetition pattern being specified: which transmission bits should be repeated, and the dotting pattern is designed such that: preferably, the transmitted bits associated with a number of input bits by said encoding process are repeated, in which method said repeating pattern is formed by: a function of the cumulative dotting intensity is found, which describes: by repeating transmission bits in a data word, which component of the input bits has been repeated, a decision function is formed depending on the accumulated dotting strength, maximizing the decision function for finding the repeating pattern.

Said communication device having a rate matching means for dotting or repeating groups of data words of a data stream to said rate matching means according to a rate matching pattern for matching a data rate of said data stream, wherein said rate matching means removes or repeats bits according to said rate matching pattern from said groups of data words by said dotting or repeating, characterized in that said rate matching means is configured such that: it implements said rate matching using a dotting pattern or a repeating pattern formed in the way described above.

Here, the systematic use of convolutional codes is used to find tentative dotting patterns, according to the use of which all bits of a dotting data word have a bit error rate corresponding to their respective significance.

The dotting pattern preferably has a dotting rate that increases from the central region of the respective data word towards both ends. In this way, the bits at the beginning and end of the respective data word to be doted are dotted more strongly, wherein this is achieved not with a uniform dotting rate but with a dotting rate which rises substantially towards the two ends of the respective data word, i.e. the spacing between the dotting bits becomes shorter and shorter on average towards the two ends of the data word. As will be explained below, however, the dotting rate surprisingly does not have to rise exactly monotonically unconditionally in the terminal direction, or in other words, the dotting pitch decreases exactly monotonically. Rather, depending on the special properties of the convolutional code used and in particular of the generator polynomial (generatorpolynomie) used, a slightly more irregular pattern can also advantageously be used.

This dotting leads to an error rate in which the individual bits are evenly distributed over the dotted data word and also has the consequence that the overall error probability is reduced.

The invention is particularly suitable for matching the data rate of a convolutionally encoded data stream and can therefore be used with advantage in a UMTS mobile radio system, wherein this relates not only to the field of mobile radio transmitters but also to the field of mobile radio receivers. The invention is not limited to this field of application but may generally be applied up to the point where the data rate of the data stream is to be matched.

Drawings

The invention is described in detail below with the aid of preferred embodiments and with reference to the attached drawings.

Fig. 1 shows a simplified block circuit diagram of a mobile radio transmitter of the invention.

Figure 2 shows an embodiment according to in HS-SCCH, part 2Bit error rate BER of each bit used for dotting, using ratio E of energy of transmitted bit to noise power density _S /N ₀ = 2dB, with R =1/3 coding. In the HS-SCCH channel is involved a so-called high speed common control channel, through which certain configuration information is transmitted, and which can be divided into two sub-regions, a so-called part 1 and a part 2. Part 1 is here transmitted first and contains the information that the mobile station needs first for processing the subsequent data channel, and part 2 contains that information which the mobile station needs later. By this division into two one can achieve that the lag caused by HS-SCCH is as small as possible, since only the first part has to be decoded before the reception of data can start.

FIG. 3 shows the ratio E of the energy to noise power density at a transmitted bit _S /N ₀ ＝-2dB In HS-SCCH, part 2, the bit error rate BER of each bit for rate matching as proposed in UMTS (technical specification 25.212v5.0.0, chapter 4.2.7 "rate matching").

Fig. 4 shows a comparison of the results which can be achieved with the inventive dotting (upper curve, circled) or the conventional dotting (lower curve, circled) with regard to the total error probability derived therefrom, wherein the probability of at least one bit of the erroneously transmitted word (so-called frame error rate) is plotted.

Figure 5 shows the basic format of convolutional codes in UMTS.

FIG. 6 shows the ratio E of energy to noise power density at transmitted bits _S /N ₀ Bit error rate BER per bit for rate matching as proposed in UMTS (technical specification 25.212v5.0.0, chapter 4.2.7 "rate matching") in HS-SCCH, part 1, when-3 dB.

Fig. 7 shows how many input bits are involved when the output bits are dotted in different output stages.

Fig. 8 shows which input bits (bit numbers) are involved by dotting.

FIG. 9 shows a table with dotting results related to the number of dotting bits.

FIG. 10 shows the signal-to-noise ratio E at the energy and noise power density of a transmitted bit _S /N ₀ Bit error rate BER per bit for dotting in the embodiment of part 1 according to HS-SCCH when-3 dB.

Fig. 11 shows a different embodiment of 8-bit dotting (48 to 40 bits) with Rate (Rate) 1/3 coding.

Fig. 12 shows a different embodiment of R =1/3, 31 bit dotting (from 111 dotting to 80 bits).

Fig. 13 shows a different embodiment of R =1/2,8 bit repetition (from 32 to 40 bits).

Fig. 14 shows different embodiments of R =1/3,6 bit repetition (from 74 to 80 bits).

Fig. 15 shows a different embodiment of R =1/2,4 bit repetition (from 36 to 40 bits).

Fig. 16 shows different embodiments of R =1/3, 14 bit dotting (from 54 to 40 bits).

Fig. 17 shows other embodiments of R =1/3, 31 bit dotting (from 111 to 80 bits). The figure can therefore also be seen as a continuation of fig. 12.

Fig. 18 shows an embodiment with R =1/3, dotted from 108 to 80 bits.

Fig. 19 shows an embodiment with R =1/3, dotted from 114 to 80 bits.

Fig. 20 shows an embodiment with R =1/3, dotted from 117 to 80 bits.

Fig. 21 shows an embodiment with R =1/2, from 52 dotting to 40 bits.

Fig. 22 shows an embodiment with R =1/2, from 46 dots to 40 bits.

Fig. 23 shows an embodiment with R =1/3, from 54 dots to 40 bits.

Fig. 24 shows an embodiment with R =1/2, from 56 dots to 40 bits.

Fig. 25 shows an embodiment with R =1/2, repeated from 36 to 40 bits.

Fig. 26 shows an embodiment from 48 dots to 40 bits.

FIG. 27 shows an embodiment from 111 dots to 80 bits.

Fig. 28 shows a rate matching specification from 3GPP technical specification 25.212v5.0.0, chapter 4.2.7 "rate matching".

Detailed Description

In the tables, rows with numbers printed in bold as a whole generally mean the respective particularly preferred embodiment, wherein, however, the quality of the further embodiments does not necessarily deviate significantly from the highlighted embodiment. However, in fig. 26 and 27, the numbers marked in bold indicate the bits at the beginning or end of the repeating pattern that are dotted or repeated by the inventive design principles described by the rate matching formulas. These bits are thus determined, and bit positions (typically up to one position) represented by bold faces can also be easily shifted by variation of the parameters within the scope of the invention.

Fig. 1 schematically shows the structure of a mobile radio transmitter 1 according to the invention, from which mobile radio transmitter 1 data or communication information, in particular voice information, is transmitted to a receiver via a high-frequency transmission channel. The components involved in the encoding of such information or data are shown in particular in fig. 1. Information provided by a data source 2, for example a microphone, is first converted into a bit string by a digital source encoder 3. The speech-coded data are then encoded by means of the channel encoder 4, the originally useful bits or message bits being encoded redundantly, as a result of which transmission errors can be detected and then corrected. In the case of the channel encoder 4 a convolutional encoder may be involved. The code rate r generated during channel coding is an important variable for specifying the respective code used during channel coding and is specified, as already mentioned, by the expression r = k/n. Here, K denotes the number of data bits, and n denotes the number of total coded bits, i.e., the number of added redundant bits corresponds to the expression n-K. A code with the above defined code rate r is also referred to as an (n, k) code, wherein the efficiency of the code increases with decreasing code rate r. So-called burst codes or convolutional codes are often used for channel coding.

The following starting point is that, as is specified by the current state of the art standardized by UMTS, convolutional codes are used for channel coding. The main difference with the block code is that in convolutional codes the individual data blocks are not encoded successively, but involve a continuous process in which each current code word of the input sequence to be encoded is also related to the previous input sequence. Irrespective of the code rate r = K/n, the convolutional code is also characterized by a so-called impact length or 'constraint length' K. The 'constraint length' illustrates that one bit affects the code word output by the channel encoder 5 by a number of few clock pulses of the k new input bits of the channel encoder 4.

As shown in fig. 5, the following convolutional code is adopted for UMTS. The image is taken from technical specification 25.212 chapter 4.2.3.1 "convolutional coding" (convolutional code).

Before the channel-coded information is transmitted to the receiver, it can be fed to an interleaver 5, which interleaver 5 rearranges the bits to be transmitted in time according to a certain format and then spreads them in time, whereby the errors that occur are distributed, usually in bundles, in order to obtain a so-called memoryless transmission channel with a quasi-random error distribution. The information or data coded in this way is supplied to a modulator 7, which modulator 7 has the task of modulating the data onto a carrier signal and transmitting it to a receiver via a high-frequency transmission channel 3 according to a predefined multiple access method.

For transmission, the coded data stream is divided into data word groups, wherein the channel encoder 4 is set in a known state at the beginning of the data word groups. At the end, each encoded data word is ended by so-called tail bits, so that the channel encoder 4 is again in a known state. With the convolutional code and the design of the channel encoder 4, it is achieved that the bits at the beginning and end of the encoded data word are better protected against transmission errors than in the center of the word. At this point, it is not important whether all of these tail bits have a known value of 0 or another value. The values of these tail bits can also be chosen arbitrarily, wherein not only the transmitter, but also the receiver must know the values to be used.

The error probability of a bit is different depending on its position in the respective data word. This effect is exploited, for example, in speech transmission in GSM mobile radio systems by placing the most significant bits at the end of the two words with the smallest probability of error. However, in the case of data transmission, data packets are generally already distorted if only one single transmitted bit is in error, which can be ascertained, for example, in the receiver by means of a so-called Cyclic Redundancy Check (CRC). So that in the case of data transmission, it is possible not to talk about important or less important bits, but to regard all bits as equally important. If an error occurs in the check word, i.e. in the data word containing the check information, it is generally no longer possible to detect the useful data correctly if only a single bit is received in error, since the received data is then interpreted in error, the check information containing information about how to encode and transmit the subsequent useful data.

In order to match the data rate of the coded data stream to the respective possible transmission rate, a rate matching is carried out before the modulator 7. In the embodiment shown in fig. 1, rate matching is performed in a rate matching unit 6b, wherein a dotting unit 6a first performs dotting according to a certain dotting pattern in order to achieve a more uniform error distribution over the data word group. The order of dotting unit 6a and interleaver 5 shown in fig. 1 should only be understood exemplarily. An interleaver may also be arranged after the unit 6 b. The interleaver 5 may equally be replaced by two interleavers before and after the rate matching unit 6b, etc.

It is therefore also the object of the invention to further optimize the dotting pattern and in particular to match it to the polynomial used for the channel encoder. The task is therefore to select the number of bit groups to be punctured or repeated in such a way that the decoding can be carried out as well as possible, depending on the convolutional code used (including the polynomial used) and the length of the word. A large number of possibilities are usually created, so that developing a good rate matching pattern purely by simulation is at least very time and resource consuming. If one were to analyze, for example, all possible dotting patterns studied for from 48-bit dotting to 40-bit dotting, there would be 48! /(8 |. 40 |) =377348994 different possibilities, which are not analytically investigated in a reasonable time.

This problem is particularly posed for short burst lengths, such as for example the check information for the UMTS extended HSDPA, and there in particular for the HS-SCCH channel (high speed shared control channel). The channel transmits configuration information that specifies how to encode the otherwise useful information transmitted over the dedicated data channel and other details, such as the spreading code used for transmission. This is relatively little data, as opposed to a data channel over which much data can be transmitted. Convolutional codes with a ratio of 1/2 or 1/3 are employed for coding in UMTS, the polynomial employed being shown in fig. 5. The exact formation of the "taps" is also called polynomial, i.e. the taps tap the lag stages of the respective output bit streams and are logically connected by an exclusive-or operation.

The invention is therefore particularly applicable to so-called HS-SCCH (high speed shared control channel).

The specification for HS-SCCH coding is given according to the current state of the art in the technical specification 3gpp TS 25.212v5.0.0 (2002-03), "multiplexing and channel coding (FDD) (program release 5)", in particular in chapter 4.6 "coding of HS-SCCH". In addition, the specification is also referred to as 25.212 for short in the present application. In chapter 4.6.6 "rate matching of HS-SCCH", it is stated that rate matching must be carried out according to the standard rate matching algorithm in chapter 4.2.7 "rate matching", which basically achieves (as far as possible) equidistant dotting or repetition.

The word length of the two parts of the HS-SCCH is 8 bits for the first part or 16 bits if the end bits (tail bits) are included concomitantly or 29 bits for the second part or 37 bits if the end bits (tail bits) are also included concomitantly, as is the current situation. Since technical specifications are still in place, it is also possible to derive additional word length by changing different parameters. Also, convolutional codes with a ratio of 1/2 or 1/3 are also contemplated. The following rate matching is especially important:

a) 32 to 40 (with code rate R = 1/2), or

b) 48 to 40 (with code rate R = 1/3), and

c) 74 to 80 (with code rate R = 1/2), or

d) 111 to 80 (with code rate R = 1/3).

Actions for determining dotting and repeating patterns

In general, it can therefore be ascertained that the dotting and/or repetition or even the repetition alone is undertaken in the case of rate matching in such a way that the overall Bit Error Rate (BER) is minimized. For this purpose, the situation in fig. 3 is first examined: where the bit error rate of each bit in a frame is scaled. The abscissa represents the exponent of the respective bit "(frame exponent)". It is clear that the bits of the first (die ersten) and last batches have a smaller bit error rate. This can be understood in connection with the convolutional code format from fig. 5: for transmission, the bits from the different lag stages D of the decoder are logically connected to one another by means of a convolutional code. The first bits are also logically connected, for example, to the bits preceding them, i.e. not present in any case. These "non-existent bits" are then set to known values, mostly zero. This is known to the receiver, which now on its part decodes the bits of the first transmission with these bits set to zero. The decoding is here very reliable, since a part of the bits is known with absolute reliability.

The same also applies to the bits of the last batch: after which artificial bits, so-called end bits or "tail bits", are inserted into the hysteresis component (Verz baby ingsgled) D of the decoder; these end bits are set to known values, mostly zero.

The bits are logically connected to each other in the central zone, and the values of these bits are not reliably known at the receiver. The probability of errors occurring is therefore greater when decoding, which is manifested in a higher bit error rate.

In this case, therefore, an envelope of the bit error rate with respect to the frame number is initially shaped convexly upward in the case of a uniform repetition or dotting. There are now different possibilities of how the envelope changes when the dotting (or repetition) is changed:

a) The envelope is essentially a horizontal line (or close to it):

this means that the bit error rates of all bits within a frame are substantially equal. This occurs, for example, by dotting on the edges, or repetition in the center, or both, and also depends on which rate should be matched.

b) Concave formation of the envelope

In this case, the dotting is so strong, for example on the edges, that the bits in the central region of the frame have a smaller bit error rate. This fact can be seen in fig. 2.

c) Irregular distribution of bit error rates with respect to frame numbers.This is not observed here in detail for reasons set forth below.

The following discussion relates to dotting. Similar considerations may be made for repetitions, or for combinations consisting of repetitions or dotting:

there are now a number of possibilities how individual bits can be dotted. For example, as already mentioned above, if one wants to analyze all possible dotting patterns studied for dotting from 48 bits to 40 bits, there is 48! /(8 |. 40 |) =377348994 different possibilities, which are not possible to investigate all analytically in a reasonable time.

The aim is to eliminate the unreasonable possibilities from the beginning. This does not occur by any repetition and/or dotting, so that alternative c) is not observed further here.

The ordering principle is shown in fig. 7. The dotting level of the respective output stage output 0, output 1, output 2 is indicated for the first 9 input bits 1-9 and the last 9 input bits n-8 to n. As can be seen in fig. 5, the output stages themselves are the respective output functions which are formed by logical connections from all input bits preceding the input bit which is observed temporally. The output stage of fig. 5 b) is here observed, i.e. a rate 1/3 convolutional encoder. For dotting with as little information loss as possible, it is proposed to first omit (dotting) bits that have little impact on the further bits. The dotting level thus indicates how many bits are affected by the dotting of the observed bits.

An exemplary manner of action to omit or dot the bit is shown in fig. 8. The first 9 input bits 1-9 are again given in the first column, and the last 9 input bits n-8 to n. The information bits referred to by dotting, i.e. the bit numbers of the information bits or input bits, of the respective output stage output 0, output 1, output 2 are shown in the following columns. As already in fig. 7, the table area (Tabellenfeld) becomes darker as the number of affected information bits rises. The bits belonging to the bright table area are therefore candidates for dotting.

Fig. 9 shows a table in which the parameters that are important when dotting near the end, i.e. the first and last bits, are shown. N input bits (information bits) and k coded bits (bits at the output stage, output bits) are observed. The number of dotting output bits (# dotting bits) is given in the first column and the number of information bits relevant to this at the input is given in the last column (cumulative Kumulative), wherein the input bits that are related to a plurality of dotting of a plurality of output bits are also counted a corresponding number of times.

In the second column it is illustrated by a sequence which output bit (bit number) has been dotted in this step. At this point, the dotting is done starting from the least significant bit in the first row and going towards the following bits in the following row. Thus, from the bits given in column 2 in rows 1 to 7, i.e. from bits 1, k,4, k-6,2, k-1, a total dotting pattern for e.g. 7 dotting bits is derived. The pattern thus comprises

bits

1,2,4, k-6, k-4, k-1, k.

The first batch of information bits 1-9, and the indication (Indizirng) of the last batch of information bits k-8 to k are located on the first row. For local reasons only-8 etc. are written instead of k-8. Entries in the columns below the information bit indication specify how strongly the relevant information bits are related by the dotting of the output bits, which are specified and thus dotted in column 2 up to the respective row. That is, how many of the dotted output bits were logically connected to the information bit. This is a scale for how strongly the information-related bits have been attenuated by dotting.

The sum of these effects is listed last in the last column (accumulation). Which is referred to herein as the cumulative dotting intensity.

The column average gives the ratio V of the sum of the last column divided by the number of information bits concerned. For example, V = (2 + 1)/(1 + 1) =1.2 is obtained for 6 dotting bits. The average dotting rate (average P-rate) is the column "average" divided by 18, the total number of xor logical connections that occur for each information bit at encoding.

The manner of action for dotting any number of bits is to make a table similar to that described above. The illustrated table can be adopted for ratios 1/3 and polynomials observed by convolutional encoders. The table may be easily determined similarly in further coding rates and/or further polynomials. With the aid of these tables, one then determines the dotting order, where the output bits that have only a minor effect on the accumulated dotting strength are first dotted. If there are multiple alternatives, then the bits that minimize the maximum of the dotting strength for each bit are preferably dotted.

For higher numbers of bits to be dotted and/or larger word lengths, it is often necessary to combine information from the table with the idea of dotting as evenly distributed as possible across the word. It is then proposed to append in the central part the targeted dotting bits generated by the generator polynomial with the least power, i.e. with the least logical connections. But at the same time note that the overall distribution of the dotting intensity in the central region of the frame does not have a significant increase.

Corresponding matters apply to the repetition, respectively with opposite signs. That is to say that the bits that have been tapped first according to the heuristic are now repeated last, and a uniform repetition in the central part is generally carried out first, preferably by a polynomial with the most logical connections. After this, bits are repeated on the edges (at the time of dotting) which have as great an influence as possible on the dotting strength of the open accumulation.

Unlike the method in which the dotting rate continuously increases in the end direction, this leads to an inherently unexpected result, because one may expect that the reliability of the coded bits continuously increases in the end direction. But it has been shown when careful inspection of the polynomial of the convolutional encoder employed that this assumption is surprisingly not correct. The coded bits are generated by polynomial-specific properties, in particular at the ends, which contribute less efficiently to the coding. But the bits do not appear to a continuously increasing extent towards the end direction but are distributed somewhat irregularly. One can further improve the coding when one aims the dotting pattern exclusively at these "weak" bits, i.e. the bits are preferably dotted.

The invention therefore uses a heuristic method which allows:

* By means of a newly specified heuristic measurement method (Metrik), the influence of the dotting/repetition of the coded bits on the basic information bits is approximately determined,

* The bits to be dotted or repeated are selected specifically and exclusively for each convolutional code,

* The number of rate matching patterns to be analyzed is strongly limited.

After a few successfully expected rate matching patterns have been found on the basis of this method, these rate matching patterns are compared with the aid of the frame error rate and the bit error rate (hereinafter referred to as bit error rate distribution) of each single information bit. The rate matching pattern can then be further refined and optimized iteratively based on the developed measurements. The bit error rate distribution of the dotless/unrepeated word is used as start-up information.

As a heuristic measure, the dotting strength S of each bit information bit i _i The number of logical connections through which an information bit having the respective output bit of the encoder passes is specified. S _i And thus positive for dotting. For the repetition, for each logical connection of n transmissions, S is specified _i，k ＝n-1。S _max Is the maximum possible dotting strength, given by the code-specific total number of existing logical connections:

a good rate matching pattern is found according to the following quality criteria:

1. selecting cumulative dotting intensities near possible minima

2. Note that the bit error rate is distributed as evenly as possible over all information bits.

Based on the generator polynomial of the code, tables are built for the beginning and end of the code word group for selecting the bits to be doted/repeated, which tables show the cumulative dotting strength of each code bit, and the associated information bits. The coded bits can thus be divided into so-called levels of accumulated dotting strength.

According to the above quality criterion, the bits to be dotted/repeated are now found by means of the tables in such a way that, firstly, for those information bits which exhibit a lower bit error rate than the other bits, the dotting strength is increased, while the cumulative dotting strength is kept small. I.e. the dotting strength is selected inversely proportional to the bit error rate of the information bits and also to purposely find bits that contribute less to the accumulated dotting strength.

The method is then iteratively applied on the basis of the first determined Pattern, so that after a small number of simulations, a rate matching Pattern (Pattern) specifically optimized for the respective convolutional code can be found.

The different possibilities of the dotting mode of the invention are shown in fig. 11 and 12, where the number of bits to be doted (counting from 1) is illustrated separately. A table is given for different numbers of information bits to be transmitted and different numbers of bits to be transmitted after rate matching.

In fig. 3, the bit error rate of individual transmitted bits of a data word is plotted against their position or position in the data word for conventional dotting with a regular dotting pattern.

This curve with the inventive dotting from pattern number 33 in fig. 12 is shown in fig. 2, which has proven to be particularly suitable in simulations. As can be seen from fig. 2, a more uniform distribution of bit error rates across the data word is achieved by employing the dotting pattern of the present invention. Since the dots are dotted less frequently in the central area of the data word than in the conventional manner of action, a smaller error probability is obtained there. The error rate now actually rises slightly towards the end, which may appear disadvantageous at first glance. However, this results in that, as already mentioned above, a particularly large number of "weak" bits are located on the edges, where dotting can be implemented quite advantageously.

The overall error rate versus the energy of the transmitted bits versus the noise power density is plotted in fig. 4 for the same case. As can be seen from fig. 4, with the present invention (lower curve, circle) an improvement of about 0.2dB frame error rate can be achieved over the conventional mode of action (upper curve, circled).

Similar improvements can be achieved with additional parameters. For example, for conventional dotting with a regular dotting pattern, fig. 6 shows the bit error rate of individual transmitted bits of a data word as a function of their position in the data word when encoded with a rate of 1/3 and 8 bits (48 to 40 bits). This is equivalent to transmitting 8 input bits. Fig. 10 shows the distribution when the dotting pattern number 3 in fig. 11 is used instead, which has also proved to be particularly suitable in simulations. It is seen that a very balanced distribution is obtained here. An improvement of about 0.2dB is also produced here. (but no curve is attached for this purpose, as it does not bring other main insights). Fig. 16 shows other preferred embodiments with 14 of the 54 bits dotting within the scope of the invention, where

patterns

3 and 4 achieve the best results.

Fig. 13, 14 and 15 show preferred repeating patterns that have also been obtained by applying the rules shown in the present invention.

The invention has been explained so far by means of use in a mobile radio transmitter. The invention can of course also be extended to mobile radio receivers, where the signals dotted or repeated in the above-described manner and method for matching the data rate have to be reprocessed (aufarbeiten) according to the respectively employed dotting or repeating pattern. In the respective receiver, additional bits are inserted into the received bit stream or two or more bits of the received bit stream are combined, for the bits which are dotted or repeated on the transmitting side. When additional bits are inserted, their information content is very unreliable, noted for these bits at the same time in the form of so-called 'soft decision' information. The processing of the received signals can be performed in the respective receivers in a reverse order to that of fig. 1, in a corresponding sense.

Other rate matching patterns obtained by using the above-described actions

The dotting patterns described so far have mainly focused on dotting in the end regions or/and repetition in the central region.

In the above-described approach for the different proposals for HS-SCCH coding in standardization, other rate matching patterns have been sought that are described. The bits to be dotted or repeated, respectively, are specified. Bits are numbered consecutively from 1 to N. The preferred mode is mentioned first, respectively, but the other modes always have similarly advantageous properties. Listing these other dotting patterns fig. 17 is therefore a supplement to fig. 12. The dotting patterns for different output bit rates are shown in fig. 18-24, respectively, while other repeating patterns are shown in fig. 25.

Approximating a preferred rate matching pattern by employing components already detailed in UMTS

The modes presented hitherto have the object of proposing the best possible selection of the bits to be dotted or repeated, wherein, furthermore, no further restrictions regarding the mode are required as a precondition. In a practical implementation it may be advantageous to observe only patterns that can be realized with as little variation as possible of existing rate matching circuits. The corresponding rate matching specification is described in the already mentioned technical specification 25.212v5.0.0 chapter 4.2.7 "rate matching". The following remarks recite part of the specification, which is a dotting or repeating step in nature, and are illustrated in chapter 4.2.7.5 "determine rate matching pattern". Excerpts from the technical specification:

using x before rate matching ₁₁ ，x ₁₂ ，x ₁₃ ，...，x _1xi Representing a bit. Where i is the transport channel number, the sequence itself is specified in paragraph 4.2.7.4 of the uplink specification and in 4.2.7.1 of the downlink. A communication connection of a communication device to a base station is understood to be an uplink and a communication connection of a base station to a communication device is understood to be a downlink.

The rules for rate matching are repeated in the program segment shown in the figure, which is executed when the conditions for implementing dotting are met.

-first setting the error value e to an initial value between the original error value and the desired dotting rate.

In a loop with the exponent m of the bit observed at the instant as the operating parameter, up to the end of the sequence, i.e. up to the exponent X _i Until now

-first setting the error value e to e-e _{Minus sign} Wherein e is _{Minus sign} Basically the number of bits to be dotted.

-then checking whether the error value e < =0

In this case, it is checked whether the bit with the exponent m should be dotted, the bit to be dotted then being set to a value δ that differs from 0 or 1.

In the case of repetition, a similar procedure is basically followed, wherein the repeated bits are then set directly after the original bits.

Bits

The bits that have been set to the value δ are then removed in the course of the next step in the dotting process, so that these bits are thus dotted.

The parameters X are selected as follows respectively _i ，e _ini ，e _{Plus sign} And e and _{minus sign} So that the desired rate matching can be achieved. Thus applying essentially to e _{Plus sign} ＝X _i ，e _{Minus sign} ＝N _p Wherein X is _i Before rate matchingNumber of bits of, and N _p Indicating the number of bits to be dotted or repeated. In principle, can be between 1 and e _{Plus sign} Optionally choose between e _ini In which a slight shift of the pattern is produced, which in some cases (rate matching after the first encounter (nesting)) is employed in order to shift the patterns appropriately to each other within different frames. The parameter i indicates in the technical specification the different transport channels. But this parameter is not relevant in this case and is therefore omitted. The following shows how the possibility of a preferred rate matching pattern of short word size in convolutional codes can be approximated by means of the already existing rate matching algorithm. In this case, it is attempted by the boundary conditions of the algorithm that the bits at the end of the code word group are preferably used during dotting, while the bits from the center of the code word group are used first during repetition. A central feature of this embodiment is that the parameter e is not used _ini Is limited to 1 to e _{Plus sign} But instead is chosen in an advantageous manner outside this range. At first glance, this choice appears to be contradictory in that the desired number of bits is no longer guaranteed to be dotted or repeated. But by advantageously matching e _{Plus sign} And e _{Minus sign} Can be achieved while still achieving the desired amount.

Setting:

X _i : number of bits before rate matching

N _p : number of bits to dot/repeat (N) _p The exponent p in (1) indicates the number of bits to be dotted, but N _p May also indicate the number of bits to be repeated).

In order to fully specify the application of the rate matching algorithm and thus the rate matching mode, the initial value of the error e must be specified separately _ini Error increment e _{Plus sign} And error decrement e _{Minus sign} Since the rate matching pattern is fully described for these parameters.

An approximation of the preferred rate matching mode is shown below with the aid of the algorithm listed in procedural release 99UMTS rate matching.

The following shows the possibility of how a preferred rate matching pattern of short burst size in convolutional codes can be approximated by means of a rate matching algorithm (data rate matching algorithm) already existing in the standard. In this case, it is attempted by the boundary conditions of the algorithm that the bits at the end of the code word group are preferably used during dotting, while the bits from the center of the code word group are used first during repetition.

Dotting

The parameters of the rate matching algorithm are selected such that the first N at the beginning of the codeword set ₀ The individual bits are dotted, for which purpose they must be applied

N ₀ ·(e _minus -e _plus )＜e _ini ≤N ₀ ·e _minus -(N ₀ -1)·e _plus (1)

As a further criterion, it is provided that the last bit of the block is also dotted, i.e. the following conditions are applied:

(N ₀ -1)·(e _minus -e _plus )＜e _ini (2)

in this case, i.e. the value of the error variable e just at the last bit, will become negative, which determines then to dot the bit. Two criteria are fulfilled, for example, by the following preferred selection of parameters:

e _plus ＝X _i -N ₀ (3)

e _minus ＝N _p -N ₀ (4)

e _ini ＝N ₀ ·e _minus -(N ₀ -1)·e _plus (5)

the special case that the bits at the start of the codeword set should not be dotted (N = 0) is also contained in these equations. Then the following applies: e.g. of a cylinder _ini ＝X _i ，e _{Plus sign} ＝X _i ，e _{Minus sign} ＝N _p 。

Selecting e according to equations (1) to (4) _ini The general implementation of (a) gives rate matching patterns that differ from those selected according to the preferred parameters of equations (3) to (5) only in the (N) th order ₀ + 1) to (N) _p -1), the exponent of the bit to be doted may be decreased by 1.

For the application example from 48-bit dotting to 40-bit, the table in FIG. 26 shows the values up to N ₀ The dotting pattern of parameter selection is preferable to that of No. 6. By e according to (1) and (2) _ini The change in value may either partially or completely lower the dot position for non-black body printing by 1.

The following table shown in fig. 27 shows in the same way the patterns resulting from dotting from 111 bits to 80 bits.

Although in this way the best dotting mode already discussed above cannot be achieved, one can still achieve with this method some improvement of the transmission quality with respect to the current state of the art specifications, wherein the changes to be made are comparatively few.

Repeat (R) to

The parameters of the rate matching algorithm are calculated in such a way that the maximum distance of the last bit to be repeated from the end of the block is ensured, i.e. it must be true that:

e _ini ＝1+X _i ·e _minus -N _p ·e _plus . (6)

furthermore, the average distance R between the bits to be repeated can be predetermined _R 。R _R Not necessarily an integer, but may be a positive rational number. Then the following applies:

thus, the quotient of R is exactly the result of R _R And a total of N repeats _p Boundary condition of each bit, e can be freely selected _{Plus sign} And e _{Minus sign} 。

If the first bit to be repeated is to be specified, the position of the first bit to be repeated (denoted b here) is to be specified ₁ ) Then, it must be applied in addition to (6)

In the formula, e _{Minus sign} Should be an integer, and b ₁ ≤X _i -N _p +1。

Preferred parameter selection results from

e _minus ＝N _p ， (9)

e _plus ＝X _i -b ₁ +1 (10)

e _ini ＝(b ₁ -1)·N _p +1 (11)

This selection with parameters is first repeated with bit b ₁ And as required, repeating N _p And (4) a bit.

The repetitive pattern generated here is also not optimal compared to the patterns already discussed above. Nevertheless, with this method, it is possible to achieve a certain improvement of the transmission quality compared to the prior art specifications, wherein the changes to be made are comparatively small. By advantageously selecting the parameter b ₁ It is achieved that the repetition is started without having to be at the starting end. I.e. repetition at the beginning is not necessary, because the bits at the beginning of the convolutional decoder already have a comparatively low error rate anyway, as demonstrated above. As occurs with this method, there is therefore a further advantage if the bits to be repeated are rather concentrated in the central direction. The disadvantage of this embodiment is, however, that the method avoids heaviness only at the starting endAgain, but less positively affects the situation on the tip. This is a penalty that must be paid for a simplified implementation.

The above criteria can of course also be combined when selecting the dotting mode. For example, one can combine two patterns described herein into one pattern by using the beginning of one of the patterns at the beginning and the end of the second pattern at the end. Furthermore, it is not important if the bits are output in a changed order while similarly matching the dotting pattern. For example, one may interchange the order of the polynomials in a convolutional encoder.

Claims

1. A method for matching data rates of data streams in a communication device,

-wherein said data stream is divisible into at least one data word containing transmission bits to be transmitted,

-wherein said transmission bits are formed from information-carrying input bits by an encoding process,

in the method, some transmission bits are removed from groups of data words of said data stream in order to match said data rate,

-wherein, by means of a dotting pattern, it is predefined: which of the transmission bits should be removed,

it is characterized in that the preparation method is characterized in that,

the dotting pattern is configured such that: 8 of the 48 bits are dotted, i.e. bits 1,2,4,8, 42, 45, 47, 48.

2. A method for matching data rates of data streams in a communication device,

-wherein said data stream is divisible into at least one data word group containing transmission bits to be transmitted,

wherein it is predefined by means of a dotting pattern which transmission bits should be removed,

it is characterized in that the preparation method is characterized in that,

the dotting pattern is configured such that: dotting 31 of 111 bits, i.e., bits 1,2,3,4,5,6,7,8, 12, 14, 15, 24, 42, 48, 54, 57, 60, 66, 69, 96, 99, 101, 102, 104, 105, 106, 107, 108, 109, 110, 111.

3. The method according to one of the preceding claims,

it is characterized in that the preparation method is characterized in that,

the transmission bits to be transmitted are transmitted over the HS-SCCH in accordance with the UMTS standard.

4. The communication device is used for the communication of the data,

-having a rate matching means (6) for dotting or repeating data words of the data stream delivered to said rate matching means (6) according to a rate matching pattern for matching the data rate of said data stream, wherein said rate matching means removes or repeats bits according to said rate matching pattern from said data words by said dotting or repeating,

it is characterized in that the preparation method is characterized in that,

the rate matching means (6) is configured such that: the rate matching means (6) performs said rate matching with a dotting pattern or repeating pattern dotting 8 of the 48 bits, i.e. bits 1,2,4,8, 42, 45, 47, 48.

5. The communication device is used for the communication of the data,

-having a rate matching means (6) for dotting or repeating data words of the data stream supplied to said rate matching means (6) according to a rate matching pattern for matching the data rate of said data stream, wherein said rate matching means removes or repeats bits according to said rate matching pattern from said data words by said dotting or repeating,

it is characterized in that the preparation method is characterized in that,

the rate matching means (6) is configured such that: the rate matching means (6) performs said rate matching with a dotting pattern or repeating pattern dotting 31 of the 111 bits, i.e. bits 1,2,3,4,5,6,7,8, 12, 14, 15, 24, 42, 48, 54, 57, 60, 66, 69, 96, 99, 101, 102, 104, 105, 106, 107, 108, 109, 110, 111.

6. The communication apparatus according to claim 4 or 5,

it is characterized in that the preparation method is characterized in that,

the communication device (1) is a mobile radio transmitter or a mobile radio receiver.